As a non-invasive and radiation-free imaging modality, electrical impedance tomography (EIT) has attracted much attention in the field of industrial measurement. However, image reconstruction with EIT is a non-linear and ill-posed inverse problem, causing it to suffer from low spatial resolution and high noise sensitivity. To overcome this problem, a point-constrained framework is proposed to guide inclusion boundary reconstruction in EIT with ultrasound point detection. The boundary reconstruction problem is formulized with an energy minimization approach. The energy function consists of the residual, point constraint, geometric and re-parameterisation terms. The residual term fits the actual EIT data by adjusting the inclusion shape and conductivity with a computable forward model. The point constraint term drags the estimated boundary toward the ultrasound detected ‘at-boundary’ points and pushes it away from the ultrasound detected ‘outside-inclusion’ points. The geometric and re-parameterisation terms regularize the boundary reconstruction problem and diffuse the point constraint energy along the inclusion boundary. According to the numerical results, the proposed method is robust to measurement noise, as well as the super parameter and initial boundary guess issues. An experimental phantom study with a water tank model further proved that the dual-modality shape reconstruction method outperforms the single-modality shape reconstruction method in both single-phase and multiphase conductivity cases.

中文翻译：

---

用于超声引导电阻抗断层扫描的点约束边界重建框架

作为一种无创、无辐射的成像方式，电阻抗断层扫描（EIT）在工业测量领域备受关注。然而，使用 EIT 的图像重建是一个非线性和不适定的逆问题，导致其空间分辨率低和噪声敏感性高。为了克服这个问题，提出了一种点约束框架，以通过超声点检测来指导 EIT 中的包含边界重建。边界重建问题是用能量最小化方法公式化的。能量函数由残差、点约束、几何和重新参数化项组成。残差项通过使用可计算的正向模型调整包裹体形状和电导率来拟合实际 EIT 数据。点约束项将估计的边界拖向超声检测到的“边界处”点，并将其推离超声检测到的“外部夹杂物”点。几何和重新参数化项使边界重建问题正则化并沿包含边界扩散点约束能量。根据数值结果，所提出的方法对测量噪声以及超参数和初始边界猜测问题具有鲁棒性。水箱模型的实验体模研究进一步证明，双模态形状重建方法在单相和多相电导率情况下均优于单模态形状重建方法。几何和重新参数化项使边界重建问题正则化并沿包含边界扩散点约束能量。根据数值结果，所提出的方法对测量噪声以及超参数和初始边界猜测问题具有鲁棒性。水箱模型的实验体模研究进一步证明，双模态形状重建方法在单相和多相电导率情况下均优于单模态形状重建方法。几何和重新参数化项使边界重建问题正则化并沿包含边界扩散点约束能量。根据数值结果，所提出的方法对测量噪声以及超参数和初始边界猜测问题具有鲁棒性。水箱模型的实验体模研究进一步证明，双模态形状重建方法在单相和多相电导率情况下均优于单模态形状重建方法。